> If the grid is the same between the timesteps, everything is alike
> step-23, easy task, but now using different grid, i am thinking of using
> Roths method, following the step-23, but the overlap between fe from
> different grid, is an obstacle i cannot pass due to the documentation of
> step-28 is a bit unclear for me.
The correct thing to do is, as you suggest, compute the right hand terms by
multiplying the solution on the old grid with test functions on the new grid.
That's awkward to implement though, as step-28 shows.
The alternative is to not change the grid very often, and then simply transfer
the previous solution from the old grid to the new grid, pretending it had
been defined on the new grid all along. This is what step-31 does, for
example (you'll have to look at the most recent svn version), and what
step-33 does all along. Take a look there!
Cheers
W.
-------------------------------------------------------------------------
Wolfgang Bangerth email: [EMAIL PROTECTED]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
